This needs
beefing up. Here is info provided by Jarno Elovirta:(IIRC, pre 1.0 versions of
FOP didn't correctly implement font selection in XSL FO files, that's the reason
for this feature existing). The PDF2 I18N allows you to say define which
characters are output with which pseudo-font, and the font mapping files define
the actual font. E.g. the configuration for English at
src/main/plugins/org.dita.pdf2/cfg/fo/i18n/en.xml
is:
<alphabet char-set="SymbolsSuperscript">
<character-set>
<!-- Copyright -->
<character>©</character>
<!-- Registered Trademark -->
<character>®</character>
<!-- Trademark -->
<character>™</character>
<!-- Service mark -->
<character>ࡈ</character>
</character-set>
</alphabet>
<alphabet char-set="SubmenuSymbol">
<character-set>
<character>➤</character>
</character-set>
</alphabet>
That
is, those specific characters are marked as "SymbolsSuperscript" or
"SubmenuSymbol" charset. The font mappings at
src/main/plugins/org.dita.pdf2/cfg/fo/font-mappings.xml then define which font to
use for them,
e.g.:<physical-font char-set="SymbolsSuperscript">
<font-face>Helvetica, Arial Unicode MS</font-face>
<baseline-shift>20%</baseline-shift>
<override-size>smaller</override-size>
</physical-font>
The
I18N processing is on by default (because it's been on before), but for most users
it's just an additional level of complexity which should be turned off and support
for multiple languages be handled in XSLT code.